opusTCv20210807+nopar+ft95-sepvoc_transformer-tiny11-align_2023-03-13.yml

release: swa-eng/opusTCv20210807+nopar+ft95-sepvoc_transformer-tiny11-align_2023-03-13.zip
release-date: 2023-03-13
dataset-name: opusTCv20210807+nopar+ft95
modeltype: transformer-tiny11-align
vocabulary:
   source: opusTCv20210807+nopar+ft95-sepvoc.spm32k-spm32k.src.vocab
   target: opusTCv20210807+nopar+ft95-sepvoc.spm32k-spm32k.trg.vocab
pre-processing: normalization + SentencePiece (spm32k,spm32k)
subwords:
   source: spm32k
   target: spm32k
subword-models:
   source: source.spm
   target: target.spm
source-languages:
   - swa
   - swh
target-languages:
   - eng
raw-source-languages:
   - swa
   - swh
raw-target-languages:
   - eng
training-data:
   swa-eng: Tatoeba-train.swa.best95.gz (4955481) 
validation-data:
   eng-swa: Tatoeba-dev-v2021-08-07, 1000
   eng-swh: Tatoeba-dev-v2021-08-07, 3
   total-size-shuffled: 1003
   devset-selected: top 1003  lines of Tatoeba-dev-v2021-08-07.src.shuffled
test-data:
   Tatoeba-test-v2021-08-07.intgemm8.multi-eng: 387/2508
   Tatoeba-test-v2021-08-07.intgemm8.shortlist.multi-eng: 387/2508
   Tatoeba-test-v2021-08-07.multi-eng: 387/2508
   tico19-test.swa-eng: 2100/56339
BLEU-scores:
   Tatoeba-test-v2021-08-07.intgemm8.multi-eng: 48.3
   Tatoeba-test-v2021-08-07.intgemm8.shortlist.multi-eng: 48.3
   Tatoeba-test-v2021-08-07.multi-eng: 49.1
   tico19-test.swa-eng: 28.9
chr-F-scores:
   Tatoeba-test-v2021-08-07.intgemm8.multi-eng: 0.62864
   Tatoeba-test-v2021-08-07.intgemm8.shortlist.multi-eng: 0.62874
   Tatoeba-test-v2021-08-07.multi-eng: 0.63415
   tico19-test.swa-eng: 0.54057